J. Appl. Entomol. 131(6), 391–399 (2007) doi: 10.1111/j.1439-0418.2007.01200.x Ó 2007 The Authors Journal compilation Ó 2007 Blackwell Verlag, Berlin Non-target organism risk assessment of MIR604 maize expressing mCry3A for control of corn rootworm A. Raybould1, D. Stacey1, D. Vlachos2, G. Graser2,X.Li2 and R. Joseph2 1Syngenta, Jealott’s Hill International Research Centre, Bracknell, Berkshire, UK; 2Syngenta Biotechnology, Inc., Research Triangle Park, NC, USA Ms. received: February 19, 2007; accepted: April 6, 2007 Abstract: Event MIR604 maize expresses a modified Cry3A protein (mCry3A), for control of corn rootworm. As part of the environmental safety assessment of MIR604 maize, risks to non-target organisms of mCry3A were assessed. The potential exposure of non-target organisms to mCry3A following cultivation of MIR604 maize was determined, and the hypothesis that such exposure is not harmful was tested. The hypothesis was tested rigorously by making worst-case or highly conservative assumptions about exposure, along with laboratory testing for hazards using species taxonomically related to the target pest and species expected to have high exposure to mCry3A, or both. Further rigour was introduced by study designs incorporating long exposures and measurements of sensitive endpoints. No adverse effects were observed in any study, and in most cases exposure to mCry3A in the study was higher than the worst-case expected exposure. In all cases, exposure in the study was higher than realistic, but still conservative, estimates of exposure. These results indicate minimal risk of MIR604 maize to non-target organisms. Key words: environmental safety, exposure, hazard, hypothesis testing, transgenic plants 1 Introduction Control of corn rootworms requires insecticides or crop rotation, and both methods can occasionally fail Corn rootworms are serious pests of maize in the USA, to prevent yield loss because of adaptation of the pest where they cause estimated annual losses of 1 billion (Rice 2003). Transgenic maize expressing insecticidal dollars in crop damage and control costs (Ostlie 2001; proteins toxic to corn rootworm offer an additional Payne et al. 2003). Corn rootworms are larvae of certain means of control (e.g. Vaughn et al. 2005). species of chrysomelid beetle of the genus Diabrotica, Syngenta has developed MIR604 maize,1 which including Diabrotica virgifera virgifera LeConte [west- expresses a modified Cry3A protein (mCry3A) for ern corn rootworm (WCRW)] and Diabrotica barberi corn rootworm control. Native Cry3A, a d-endotoxin Smith and Lawrence [northern corn rootworm produced by Bacillus thuringiensis subsp. tenebrionis, (NCRW)] [European and Mediterranean Plant Protec- has little or no activity against WCRW or NCRW tion Organization (EPPO) 2004]. D. v. virgifera is an because of the limited processing of protein in the guts alien invasive species in Western and Central Europe. It of these insects; mCry3A contains introduced cathep- was introduced from the USA at least three times in the sin G-recognition sites, which allow activation of the 1990s and early 2000s (Miller et al. 2005), and popula- protein by gut proteases in WCRW and NCRW. tions reached economic thresholds soon after the first mCry3A has greatly increased toxicity to these insects, introduction (Kuhlmann and van der Burgt 1998). and expression of mCry3A in maize provides protec- Corn rootworms feed on the root systems of maize tion against WRCW and NCRW (US Patent no. seedlings. Moderate pruning of roots by corn root- 7,030,295). This paper is an assessment of the risk to worms can lead to damage during water shortages non-target organisms of MIR604 maize. because of the inability of the maize plant to adequately translocate water and minerals from the roots to the rest of the plant. Maize plants suffering from moderate 2 Problem formulation and hypothesis testing to severe root pruning are also susceptible to lodging during rain and wind storms (Levine and Oloumi- Assessment of the safety of transgenic plants expres- Sadeghi 1991). Maize ears on lodged plants may be sing insecticidal proteins relies on the principle of underdeveloped or not be harvestable because of inaccessibility to harvest equipment. 1MIR604 maize will be sold as AgrisureTM RW hybrid maize. 392 A. Raybould et al. comparative risk assessment: the transgenic plants are tests for detecting effects of insecticidal proteins it compared with non-transgenic, near-isogenic counter- should be remembered that endpoints are used to parts that are considered to have no unacceptable predict effects on organisms in the field; an endpoint effects on non-target organisms, and any differences may be highly sensitive and have high power to detect are evaluated (e.g. Kuiper et al. 2002). If plant an effect, but if it cannot be interpreted biologically characterization data show that the only ecotoxicolog- then it has no value for risk assessment and will ically relevant difference in the transgenic plant is needlessly trigger further evaluation. Power to detect expression of the intended insecticidal protein, which is an effect is important, but there must be a hypothesis the case for MIR604 maize [Food Standards Australia to link the effect to changes in the assessment endpoint New Zealand (FSANZ) 2006], the risk assessment (Raybould 2006), which in this case is the abundance should seek to predict the likelihood of harmful effects and diversity of non-target organisms potentially of the protein to non-target organisms. Such predic- exposed to mCry3A via MIR604 maize. tions are made from tests of risk hypotheses. However well-chosen, there is always the possibility The exposure of non-target organisms to an insec- that that the representative indicator species are less ticidal protein is estimated from plant expression sensitive or have lower exposure than some of the data, the diets of non-target organisms, the rate of species for which they are surrogates. Therefore, to degradation of the protein in soil, and any other increase confidence in the risk assessment, representa- relevant environmental fate data. The worst-case tive indicator species should be tested at concentra- concentration of the insecticidal protein to which tions of the insecticidal protein in excess of the EEC. If a particular non-target organism may be exposed is under these conditions the insecticidal protein has no called the expected environmental concentration adverse effects on representative indicator species that (EEC). Some species will not be exposed to the have potentially high sensitivity and exposure, there is protein, and the safety of the transgenic plant to these high confidence of minimal risk of the transgenic plant non-target organisms can be demonstrated without to all non-target organisms. toxicity testing; in effect, the risk hypothesis under test Tests using purified proteins are more rigorous tests is that the EEC is not greater than zero. Many of risk hypotheses than studies using plants expressing non-target organisms will be potentially exposed to those proteins, because test species can be exposed to the protein (EEC >0), and for these species a high concentrations of the protein under controlled conservative risk hypothesis is that the no observable conditions (Romeis et al. 2006; Garcia-Alonso et al. effect concentration (NOEC) is greater than or equal 2006; Raybould 2006). When using a protein from to the EEC. Data on the toxicity of the insecticidal a source other than the transgenic plant, it is necessary protein are required to test this hypothesis. to show that it is equivalent to the protein expressed in It is not possible to test all species for which the EEC the plant. The so-called Ôbridging studiesÕ test for of the insecticidal protein is greater than zero. There- differences in parameters such as molecular weight, fore suitable representative indicator species are tested, glycosylation, cross-reactivity to antibodies and bioac- which act as surrogates for species not tested (e.g. tivity against sensitive species. Garcia-Alonso et al. 2006). Confidence in the risk Laboratory studies using plant material will be assessment is strengthened by increasing the rigour subject to similar constraints as studies using proteins with which the risk hypothesis is tested, and therefore in terms of the number of species that can be assessed: the best representative indicators are those species it will always be necessary to extrapolate results from most likely to reveal toxicity of the protein at concen- representative indicator species to species that were not trations close to the EEC (Raybould 2006). These tested. However, it is difficult to attain concentrations species could be taxa closely related to the target pest, of the protein greatly in excess of the EEC using plant and hence likely to have lower NOECs than most of material, and thus the ability to extrapolate from the species for which they are surrogates, or species indicator species to untested species is reduced. Field that have high exposures, and hence likely to have studies can assess the effects of exposure on many higher EECs than most of the species for which they species, but they suffer from lack of control of are surrogates. environmental variables, reducing the power to detect The choice of representative indicator species is not effects. In addition, correlations between confounding the only consideration when testing for hazards of environmental variables and the presence or absence of insecticidal proteins; the developmental stage of the the insecticidal protein may make it difficult to representative indicator, the endpoints of the study and interpret any effects that are detected (Raybould 2006). the length of exposure can also have strong influences From the discussion above, a list of criteria for on the ability to rigorously test risk hypotheses. In suitable representative indicator species for hazard general, young animals are more sensitive than older assessments of insecticidal proteins can be formulated: animals (e.g. James et al. 1999; Betz et al. 2000; • Close taxonomic relatedness to the target pest.